[XeTeX] Issue with CJK in pdf build

[Chris Jones]

On Wed, Nov 18, 2009 at 10:56:24PM EST, maxwell wrote:
> Wilfred van Rooijen <wvanrooijen at yahoo.com> wrote:

Actually, I wrote the following, not Wilfred:

> > What I don't understand is that the dblatex manual apparently states
> > that mixing 'languages' (scripts, rather) is not possible.
> 
> I'm getting into this conversation late (I'm on vacation), and I haven't
> read all the previous msgs.  But--if the dblatex manual says that, then the
> manual is out of date wrt the program.  That might be because dblatex was
> originally set up to produce only LaTeX output, i.e. with an 8-bit
> encoding, for which mixing languages might be a problem.  As it turned out,
> that was not a real limitation, and I was able to get XeTeX output from it
> by a kludge a couple years ago.  The author of dblatex, Benoît Guillon,
> quickly modified his program so as to produce valid XeTeX output using a
> command line parameter to specify the target, so I no longer needed the
> kludge.

Here's what I was writing about:

  http://dblatex.sourceforge.net/doc/ch03s07.html

I have a feeling that this may be misleading only for the likes of me. I
now suspect that 'lang' is a DocBook attribute as in <article lang=en>
and this is a warning against using several of these attributes in the
same document. So, the author would appear to be talking about DocBook
input restrictions rather than TeX/LaTeX (?)

> We have used dblatex + XeLaTeX to produce grammars where English is the
> language of description, and the examples etc. are in Bengali, Urdu, and
> now Pashto.  We use a small xetex command file to set the main font and to
> assign commands to use fonts for other languages, and to make a few other
> adjustments--basically a small style file.  And it all works together quite
> well.

Which is really wonderful. I'm avidly reading Michiel's Japanese Grammar
(sorry, I don't have the exact title in front of me) and finding it
considerably more interesting than digging into the quirks or my tool
chain :-)

> Since xetex needs to know what font to use for particular characters,
> unless you have a font that covers all the characters in your document you
> need to mark up the text for which font to use.  I won't go into the
> details, but in our case, that's made more complex by the fact that Urdu
> and Pashto are written in a right-to-left script.  For Chinese, that should
> not be an issue.  You can just run a script that selects contiguous
> sequences of Chinese text (and possibly other characters, like punctuation)
> and tags them in xetex for the Chinese font (i.e. the command to use the
> Chinese font, as per the previous paragraph).

> Since it sounds like some of your difficulties are with the use of
> dblatex, you might want to post some of your questions to the dblatex
> mailing list.

I did post something there at a time where I knew so little that I
probably didn't  make much sense and Benoît was going to look at it over
the weekend. Well.. I guess that now the problem is solved, I can post
the solution. ;-)

CJ